Fast Performance Data Guard Tips Oracle Database Tips by Donald BurlesonDecember 9, 2015

Oracle Data Guard- Performance Tuning of Data Guard Configuration

So far in this book, various "Do's and Don'ts" about Oracle instance implementation in the creation of a well-tuned Data Guard configuration have been presented. The following is a list of consolidated points used in this section as a set of guidelines to achieve better performance from Data Guard:

Configuring Data Guard for High Performance

• LAN and WAN topologies should provide a suitable bandwidth for timely delivery of archived redo logs to remote destinations. Consider a Gigabit Ethernet for LAN applications and a T-3/E-3 topology for WAN applications.
   

• Standby databases participating in high data protection modes such as MAXIMUM PROTECTION should be carefully chosen. A Oracle instance should not be connected over a WAN to participate in maximum data protection. This will significantly impact the performance of the primary database.
   

• Consider using a cascaded standby database configuration for standby sites connected over WAN. In this configuration, a remote archival destination on a LAN will receive archived redo log files from the primary database and distribute to standby sites connected over WAN.  This will reduce the load on the archiver process of the primary database.
   

• Any reporting requirements and backup activities should be offloaded to a suitable standby database.
   

• On the remote archiving destination, use a fast disk and RAID technology to improve the I/O of the log transfer service.
   

• Consider using Secure Shell port forwarding to compress the archived redo logs before sending to the remote destination. Secure Shell port forwarding may not always result in better performance; therefore, this option must be evaluated in conjunction with the data protection mode of the Data Guard configuration.
   

• Regularly monitor the log apply process on standby databases, especially the logical standby databases. The logical standby databases are more likely to be hit by performance problems than a physical standby database.
   

• Physical standby databases should have redo logs configured to minimize the time required for role transfer during switchover or failover.